JPH0354498B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION In order to eliminate undesired oscillations that persist and are a nuisance within a closed loop system, the present invention incorporates a variable gain amplifier in a first path forming part of the loop of the system. This invention relates to an integrated undesired oscillation prevention circuit.

Among closed-loop systems in which undesired oscillations may persist are, for example, systems for controlling electrical, mechanical or physical quantities. When designing such a system, it is necessary to pay appropriate attention to the gain and phase conditions of the control loop to avoid sustained oscillations within the control loop. In a given system, there is no guarantee that large amplitude and potentially dangerous self-oscillations will not occur due to unforeseen circumstances such as parasitic coupling or sudden perturbations. Another type of closed-loop system is an electroacoustic system, which includes, for example:
It has a microphone and a speaker (loudspeaker) that are coupled in an arbitrary manner by an electrical circuit, and when these two transducers are coupled by an acoustic circuit, oscillations occur at high amplitudes, which is unacceptable. This results in an electroacoustic loop that causes howling. This phenomenon of howling is known as the Larsen effect, and can occur, for example, in audio playback devices and loudspeaker telephones.

A conventional method generally used to prevent oscillation in an electroacoustic loop is to provide this loop with at least a variable amplifier or attenuation circuit, and to control these according to different criteria so that the loop gain is 1 or less. It's a method. The method adopted for public address telephones is to use two envelope detectors to check whether the audio signal is present in the speaker path or the microphone path, and to increase the gain in the effective path and detect the gain in the other path. This is a method of reducing gain. Performing this kind of gain modulation in both paths during a conversation would be extremely unpleasant for the listener, and in addition, this type of method is not effective because the gain control does not substantially depend on the acoustic coupling coefficient. isn't it.

Furthermore, another method disclosed in French Patent Application No. 2461412 employs two signal envelope detectors in the two paths and a single amplifier in the speaker path, with the gain of this ,
The difference signal between the output signal of the envelope detector in the microphone path and the output signal of the envelope detector in the speaker path is changed by a signal that is given a predetermined weight. In this method, if the distance between the loudspeaker and the microphone is greater than a predetermined minimum distance, the gain of the amplifier is a constant ratio to the acoustic coupling coefficient; Thus, oscillation can be prevented. In addition, since an envelope detector is used, the received voice signal can be made dependent on the voice signal, particularly the voice signal of the speaker speaking into the microphone.

The object of the present invention is to provide another type of loop oscillation prevention device that eliminates the drawbacks and limitations of prior methods. The invention relates not only to electroacoustic loops, but also to all other types of closed loop systems, such as control systems.

The invention provides that when a loop oscillation condition occurs, the loop oscillation is passed through a second path having at least a portion distinct from the first path and the oscillation is routed by a regulator to the remainder of the loop. In part, this is based on trying to bring it to a certain level that is low and does not become a nuisance.

According to the circuit according to the invention, the second path is formed between the input terminal and the output terminal of the first path and has at least a part separated from the first path, and over the entire frequency band in which oscillations may occur. means for maintaining the gain of the second path higher than the gain of the first path; supplying an input terminal of a linear amplifier controlling at least one variable gain amplifier forming part of a second path, the signal at the input terminal of the linear amplifier forming a certain value of said signal formed in said second path; The feature is that the configuration is such that a constant value is maintained from .

In an embodiment of the invention, the second path is completely separated from the first path and the second path is provided with a variable gain amplifier producing an output signal that is applied to the input terminal of the linear amplifier. controlling the variable gain amplifier of the second path by an amplifier, and at the same time, the variable gain amplifier provided in the first path near an input terminal shared by the first and second paths; and the first path. It is preferable to control both or either one of the variable gain amplifiers provided near the output terminal shared by the first and second paths.

In practicing the present invention, the first and second paths further include a common variable gain amplifier located near an input terminal shared by both paths and a common variable gain amplifier located near an output terminal shared by both paths. The linear regulator can control one or the other of the variable gain amplifiers shared by both paths. variable gain amplifier.

Advantageously, it is possible to clearly define the frequency of non-nuisance oscillations occurring in the loop closed by the second passage, which is preferably determined in a different part of the second passage than the first passage. This can be done by connecting a narrow passband filter in front of the point of tap of the signal to be applied to the input terminal of the linear regulator in order to selectively increase the gain of the second path.

In addition, in order to satisfy the gain conditions for both paths, another filter is included in a portion of the first path that is different from the second path to increase the gain in the same narrow pass band as the filter in the second path. Alternatively, such a filter can be provided before the common input terminal, after the common output terminal, or in a common part of both paths.

The circuit according to the invention may be advantageously employed in a public address telephone, in which case the coupling circuit for coupling the telephone to the telephone line is included in a different part of the first passage than the second passage. In the loop, hereinafter referred to as auxiliary loop, which has an acoustic path between the loudspeaker and the microphone of the telephone and which is closed by a second path, the regulator is used to At the same time as the oscillations of such modulated amplitude persist, there is a failure in the coupling circuit that couples to the subscriber's telephone in the electroacoustic loop, hereinafter referred to as the main loop, or an acoustic coupling occurs in the distant subscriber's telephone. nuisance or undesired lassen oscillations that may otherwise occur are avoided.

When the circuit according to the invention is used in a hands-free type loudspeaker telephone, automatic switching of the gain of the transmitting and receiving paths of the telephone can be achieved in a very simple manner. In this regard, it should be noted that in many conventional hands-free telephones, including the telephone disclosed in French Patent Application No. 2376576, in order to prevent the occurrence of this The advantage is that it uses a gain switching method. This type of telephone detects the level of the audio signal in both the transmitting and receiving paths, determines the path with the higher audio signal level, and increases the gain of this path while reducing the gain of the other path. In this case, both of these gain modulations are carried out in a complementary manner so that the total gain of the loop, where there is a risk of generating Lassen oscillation, is 1 or less.

In a hands-free type telephone equipped with a circuit according to the invention, there is no need for audio gain switching to prevent the occurrence of undesired Lassen oscillations, but manual gain switching or audio signal switching is not necessary for other reasons. Therefore, it may also be beneficial to perform switching control. For example, when using a hands-free telephone in a very noisy situation, the gain of the receiving link is increased to ensure a good reception condition, and
It may be necessary to speak louder or closer to the microphone, but it may be important to reduce the gain of the transmit link. In that case, this gain switching may be controlled manually or directly by the speaker. In this second assumption, when the speaker is not speaking at a hands-free telephone, the gain of the receive link must be increased to improve listening conditions, and the gain of the transmit link must be decreased. It is necessary to make sure that the noise around the area where this phone is placed cannot be heard. On the other hand, when a speaker speaks to a hands-free telephone, the gains of both the transmitting and receiving links can be set to normal values.

It has also been found that the circuit according to the invention makes it possible to use very simple means for performing gain switching in both the transmitting and receiving paths of the telephone.

According to a variant of the invention, in the case of a telephone incorporating a circuit according to the invention and comprising in the first path a variable gain amplifier controlled by a linear regulator and arranged in the receiving path. is a microphone and
further comprising, in the transmission path between the input terminals shared by the individual portions of the first and second paths, another variable gain amplifier having the function of switching the gains of the first and second paths; the means provided for maintaining the gain of the passageway higher than the gain of the first passageway, the electroacoustic loop closed by the second passageway being a source of undesired oscillations of low amplitude; and may be configured to adjust and determine an opposite gain change in the receive path by the change caused by the gain of said further variable gain amplifier.

In another variant, the loudspeakers are provided in separate parts of the first and second paths when the transmission path includes a variable gain amplifier included in the first path and controlled by a linear regulator. Therefore, a transmission path between the output terminal and the shared output terminal is provided with another variable gain amplifier having a function of switching the gains of both the first and second paths, and the gain of the second path is changed from the gain of the second path to the gain of the first path. said means provided for maintaining the gain higher than the gain of said second passage are adjusted such that the electroacoustic loop closed by said second passage is a source of undesired oscillations of low amplitude; The change caused by the gain of the variable gain amplifier can be configured to determine an opposite gain change in the transmit path.

A variant of the circuit of the invention employing a variable gain amplifier controlled by a linear regulator and shared by first and second paths, intended for gain switching and for either the transmit path or the receive path. By controlling the gain changes in the amplifiers in the transmitter and receiver, the gain changes in both the transmitting and receiving paths can be made accurately and complementary.

In the above two variants of gain switching,
2 using a detector of the audio signal from the microphone
It is possible to control the gain change of the gain-switched amplifier to provide automatic gain-switching of the two paths.

The invention will be explained below with reference to the drawings.

FIG. 1 shows an electroacoustic loop formed by, for example, a public address telephone. This phone has telephone line 2
a coupling circuit 1 for coupling the signal to a transmitting path 3 and a receiving path 6, the transmitting path being provided with a microphone 4 and an amplifier 5, and the receiving path being provided with a receiving amplifier 7 and a speaker 8. ing. A certain degree of acoustic coupling occurs between the transducers of the loudspeaker 8 and the microphone 4, depending in particular on the distance and direction between them; This coupling may change for a type of telephone commonly referred to as a "receiving" telephone. This coupling can be expressed as a coupling coefficient 1/β<1. Here β represents the attenuation of the acoustic power transmitted by the loudspeaker and reaching the microphone.

Furthermore, since there is an unavoidable failure in coupling circuit 1,
The signal appearing on the transmitting path 3 of the telephone cannot be completely retransmitted via the telephone line 2, and a portion of this signal enters the receiving path 6. Another undesirable problem is caused between the transmitting and receiving paths of the subscriber telephone shown in the figure by the path with the telephone line 2 connected to the distant subscriber loudspeaker where there is acoustic coupling between the loudspeaker and the microphone. There is a risk that binding may occur. Whatever the cause of the undesired coupling between the transmitting path 3 and the receiving path 6, a standard ratio between the signals appearing on these transmitting paths 3 and the signals 6 appearing on the receiving path 6 is established between the transmitting and receiving paths of the telephone. It is possible to provide a device 9 with a constant gain G ₁ . In this case, for example, if a circuit 10 such as an amplifier is arranged in series with the coupling circuit 1 and the gain G ₁ includes the gain of this circuit 10, then
The overall definition of this device 9 remains unchanged.

The acoustic coupling between the loudspeaker 8 and the microphone 4 and the undesired coupling between the transmit path 3 and the receive path 6 of the telephone form an electroacoustic loop whose gain is equal to or greater than 1. If it is large, oscillation may occur in this loop.
The exact frequency of these oscillations is unknown, but lies in the passband of the components included in this loop, or approximately 300-4000 Hz. The amplitude of these oscillations is limited only by the saturation of the loop components.

These undesired loop oscillations make it virtually impossible to use the telephone, so in order to avoid this oscillation, a variable gain amplifier 11 is included in the receiving path 6 of the telephone, and the envelope of the signals in both the transmitting and receiving paths is It is known to control this gain g ₁ from the signal.
This gain control method is insufficient when acoustic coupling is strong, making the gain g ₁ dependent on the effective audio signal.

The present invention aims at preventing oscillations in a completely different way, thereby avoiding the above-mentioned drawbacks of the prior art. According to the invention, the first path of the loop containing the device 9 and the variable gain amplifier 11
Another variable gain amplifier 12 at terminals A and B of C ₁
Connect the second passage C ₂ containing. This amplifier 12
The gain g ₂ is controlled by a linear regulator 13, which receives as an input signal the signal W produced at the output terminal of the amplifier 12 and outputs a gain control signal.
vg ₂ , thereby maintaining the level of the signal W at a constant value above a certain value of the input signal x of the amplifier 12.

FIG. 2 is a characteristic diagram showing this output signal W as a function of the input signal x of the amplifier 12 controlled in this way. When the value of the input signal x becomes x _p , the value of the output signal becomes W _p . Up to this value x _p , the amplifier gain g ₂ is constant and is the maximum value g ₂ M = W _p determined by the slope of the straight line OC. /x is equal to _p . When the value of x _p occurs, the level of the output signal W becomes constant and equal to W _p . For any value x of the input signal, the operating point of the amplifier is M, and its gain g ₂ =W _p /x decreases with increasing x. It should be emphasized here that because of the linear regulator 13 associated with the amplifier 12, this amplifier does not introduce any nonlinearity into the second path.

The amplifier 11 of the first path C ₁ is controlled by a gain control signal Vg ₁ provided by a regulator 13 such that its gain g ₁ is responsive to the gain g ₂ of the amplifier 12 of the second path. That is, for example, the gain g ₁ of amplifier 11 can be made equal to or proportional to the gain g ₂ of amplifier 12 at each instant. If these two amplifiers 11 and 12 have the same configuration, the control signal
Vg ₁ and Vg ₂ may also be made equal and these signals may be derived from the same output terminal of the regulator as shown.

Further, one or both of the two paths is provided with means for making the gain of the second path _C2 greater than the gain of the first path _C1 in all frequency bands where oscillations may occur. In the embodiment shown in FIG.
G ₂ with a gain of 1/A 2 provided before and after the variable gain amplifier 12 of the second path C ₂ and preferably connected between the output terminal of the amplifier 12 and the common terminal B of the two _paths . is shown by an amplifier 14 associated with an attenuation circuit (attenuator) 15. Using the notation described above, this gain condition for the two paths can be written as G ₂ .1/A ₂ .g ₂ >G ₁ .g ₁ (1).

What should be noted here is that this condition (1) should hold true for all possible frequencies of loop oscillation, that is, for all frequencies in the band of 300 to 4000 Hz in this embodiment.

If the two amplifiers 11 and 12 are equivalent and controlled by the same signal, then condition (1) becomes G ₂ ·1/A ₂ >G ₁ .

In order to explain the method of the present invention and its ability to prevent undesired oscillations, the speaker 8
and its amplifier 7, microphone 4 and its amplifier 5
It is useful to define the gain G of the electroacoustic circuit shown by the dashed line 16, with the acoustic path defined by the coupling coefficient 1/β between the loudspeaker and the microphone.

This gain G is expressed as a signal v supplied to the amplifier 7.
and the signal u generated from the amplifier 5.

Let p be the acoustic power up to the output of the speaker, and let q be the acoustic power after the microphone. In that case, clearly p/q=1/β (2). In order to introduce the loudspeaker conversion factor, a nominal acoustic power p _p supplied by the loudspeaker 8 can be defined. This power p _p is generated by the signal v _p applied to the input terminal of the amplifier 7 . amplifier 7
The conversion coefficient of the combinational circuit formed by the speaker 8 and the speaker 8 is [p _p /v _p ]. To the extent that this combinational circuit has linearity, it can be expressed as p=v [p _p /v _p ] (3).

In order to introduce the conversion factor of the microphone, we measure it directly in front of this microphone and obtain it by a loudspeaker located at a nominal distance expressed by the coupling coefficient 1/β and producing an acoustic power Kp _p (K is a constant). Let q _p be the acoustic power generated. The signal u _p occurring at the output of the amplifier 5 corresponds to this acoustic power q _p =Kp _p /β _p .

The conversion coefficient of the combinational circuit consisting of the microphone 4 and the amplifier 5 is [u _p /q _p ]=[u _p /Kp _p ·β _p ].

Within the range where this combinational circuit exhibits linearity, it can be expressed as u=q[u _p /Kp _p ]·β _p (4).

Considering relational expressions (2), (3), and (4), the gain G of the electroacoustic circuit 16 becomes G=u/v=β _p /β・u _p /v _p・1/K (5) .

According to conditional expression (1), the second passage between terminals A and B
Since the gain of C ₂ is greater than the gain of the first pass C ₁ ,
Oscillations can only exist in the auxiliary loop formed between the electroacoustic circuit 16 and the second passage _C2 , which circuit 1
6 and the main loop formed by _the first passage C1. First, it is assumed that no audio signal generated from the microphone 4 or an audio signal from the telephone 2 exists in this auxiliary loop. When the gain of this auxiliary loop is less than 1, no oscillation occurs, that is, when _G.G.sub.2.1 / _{A.sub.2.g.sub.2} < ₁ (6), no oscillation occurs. In this inequality, the gain g ₂ of amplifier 12 is the maximum value as defined previously.
It is necessary to give g ₂ M. According to equation (5), the gain G
Since is proportional to the coupling coefficient 1/β between the speaker and the microphone, it can be easily said from inequality (6) that there is no oscillation if the coupling coefficient is lower than a predetermined value.

When the coupling coefficient between the speaker and the microphone increases, the gain G increases, and when the coupling coefficient exceeds a predetermined value, equation (6) is no longer satisfied. Therefore, oscillation occurs in the auxiliary loop formed by the circuit 16 and the second path _C2 . The linear regulator 13 reduces the gain g ₂ of the amplifier 12 so that the amplitude of these oscillations at the output of said amplifier remains at the value W _p . In this case, the output signal of the attenuation circuit 15
The value of W _p /A ₂ is low enough such that all of the loop components, especially amplifiers 5 and 7 of circuit 16, operate in linear mode. In this oscillation mode, the overall gain of _the loop is set to a value of 1 and can be expressed as _G.G.sub.2.1 / _{A.sub.2.g.sub.2} =1 (7).

Since the components of the circuit 16 operate in a linear mode, g ₂ = β/β _p [A ₂ / G ₂・v _p /u _p・K] (8) is easily obtained.

From this equation (8), it can be seen that for this oscillation mode of the auxiliary loop formed by the circuit 16 and the second path C ₂ , the gain g ₂ of the amplifier 12 is proportional to the acoustic attenuation coefficient β between the loudspeaker and the microphone. Set by value. This attenuation coefficient β is substantially proportional to the distance between the two transducers.

The gain of the first path amplifier 11 in response to the gain g ₂ is therefore substantially proportional to the acoustic attenuation coefficient β, ie the distance between the loudspeaker and the microphone. This gain g ₁ changes in the same direction as the gain g ₂ , making it possible to satisfy condition (1) under all circumstances. If this condition (1) is satisfied,
Since the total gain of the main loop formed by the circuit 16 and the first path C ₁ that closes it is always less than 1, this first path C ₁ is connected to the electroacoustic circuit 16.
does not act in a manner that causes large amplitudes or undesired lassen oscillations at the terminals.

Thus, according to the circuit or device according to the invention, when the acoustic attenuation coefficient becomes sufficiently small to create conditions for oscillations in the loop, these oscillations are forced to pass through the second path _C2 . . The reason is that especially the gain g ₁ of the amplifier 11 in the first path C ₁
is responsive to the gain g ₂ of the second path amplifier 12. In this second path, the loop oscillations are controlled so that these oscillations do not become a nuisance in the electroacoustic circuit. That is, these oscillations can be made inaudible and do not saturate amplifiers 7 and 5.
Therefore, even when the amplitude of oscillation at the output terminal of amplifier 12 takes a predetermined maximum value W _p , the damping coefficient A ₂ of circuit 15 is always increased to minimize the amplitude of oscillation at the input terminal of circuit 16. If this is not a nuisance, it is possible to increase the gain G ₂ of the circuit 14 correspondingly so that condition (1) is satisfied. The signals generated at the output terminal B by the oscillation around the second path _C2 all have the same value V _L =
W _p /A ₂ .

Next, we will discuss the case where the first path C ₁ and the electroacoustic circuit 16 are effective or desired signal sources. First, it is assumed that an audio signal generated from the microphone 4 is directed from the terminal A to the first path _C1 . In this case, as in the case described above, loop oscillations may exist in the second path _C2 , unaffected by the audio signal from the microphone. Output terminal B
, the amplitude of these loop oscillations is extremely small and is superimposed on the voice signal coming from the telephone line 2 via the output terminal of the first path _C1 . A person listening in front of the speaker 8 is unable to open these loop oscillations, which have extremely small amplitudes. In the case of the assumptions explained above, the gain g ₂ of the amplifier 12 and thus the gain g ₁ of the amplifier 11 is not influenced by the audio signal from the microphone 4 and depends only on the acoustic coupling coefficient β.

In fact, if precautions are not taken, microphone 4
A portion of the signal from the loop oscillation is supplied to the second path _C2 and disturbs the amplitude adjustment of the loop oscillation by the regulator 13. In that case, the gain g ₂ of the amplifier 12 and thus the gain g ₁ of the amplifier 11 depends not only on the acoustic coupling coefficient but also on the amplitude of the audio signal from the microphone 4. In order to eliminate such drawbacks, a filter 17 is provided in the second path _C2 upstream of the amplifier 12 as shown by the broken line in the figure, and the gain of the second path is adjusted to a specific part of the frequency band where oscillation is likely to occur. is maintained higher than the gain of the first path. in this case,
The gain G ₂ of the second path will include the gain of amplifier 14 and filter 17 .

In the first path _C1 , for example, if the gain for frequencies above 3000Hz is higher than the gain for frequencies below 3000Hz and oscillation is likely to occur above 3000Hz, a high-pass filter can be used as the filter 17. Increase the gain of the second path for high frequencies. In this way, condition (1) can be easily satisfied, and the negative influence of the audio signal on the adjustment of loop oscillation is also reduced. Although these loop oscillations occur at frequencies above 3000Hz, they are still not sufficient.

It is further advantageous if this filter 17 is a bandpass filter with a very narrow band. Such a filter selectively increases the gain G ₂ in its passband, and the gain of the auxiliary loop formed by the circuit 16 and the second path C ₂ takes a positive value in this passband. without regard to the acoustic coupling of the speaker and microphone.
The oscillation in this auxiliary loop is made to occur approximately at the center frequency of this passband. At the same time, the second path audio signal is of a significantly lower level compared to the loop oscillations and does not substantially disturb the amplitude adjustment of these oscillations.

What should be noted here is that the narrow band filter 1
7, condition (1) can be satisfied much more easily and the loop oscillation can be ensured to pass only through the second path _C2 . To meet this condition, the gain of the second path for the center frequency of the narrow band of filter 17 must be higher than the gain of this first path for all frequencies in the band of the first path (for example, 300-4000 Hz). is sufficient. This condition (1) can be more easily satisfied by providing a filter 18 that increases the gain in the same narrow band as the filter 17, as shown by the broken line in the first path. In that case, the first path gain G ₁ includes the gain of device 9 and filter 18 . This filter 18 has a gain G ₁ in this passband.
is selectively increased, so that the gain of this first path C ₁ takes a maximum value in the narrow passband of this filter 18. In order to satisfy condition (1) and thus to ensure that the loop oscillations pass along the second path _C2 , for the common center frequency of the passbands of filters 17 and 18, the second path C It is sufficient that the gain of _C2 is higher than the gain of the first path _C1 . As can be seen here, since the filter 18 is narrow band, it will cause virtually no perturbation to the audio signal. Also, on the one hand, the amplifier 14 and the filter 17, and on the other hand, the amplifier 10
and filter 18 may be in the form of two selection amplifiers. Furthermore, a filter 18 is installed in the first passage _C1.
Instead of using a filter, a filter as shown by the broken line 19 in the figure may be provided in the path between the output terminal of the amplifier 5 and the input terminal A, which is shared by the two paths. Since this filter 19 increases the gain in the same narrow band as filter 17, condition (1) is automatically satisfied and the loop oscillations at the center frequencies of the two filters 17 and 18 are passed through the second path. In addition, the filter 17 in the second passage may be
In the above-mentioned variant in which the high-pass filter increases the gain G ₂ for frequencies higher than the above, it is also possible to use the filter 18 or the high-pass filter 19 in order to prevent the audio signal from being distorted to a considerable extent. However, it is clear that there is a slight attenuation difference between the high frequency bands that are transmitted and the low frequency bands that are attenuated.

Several different variants of the circuit arrangement according to the invention are also possible and will be explained using the corresponding circuit diagrams. In these figures, components that are the same as those shown in the circuit diagram of _FIG . 14 is an amplifier with gain G ₂ , 15 is an attenuation circuit with gain 1/A ₂ provided in the second path C ₂ , and ₁₃ is a linear amplifier. Also, amplifiers 11 and 12 having gains g ₁ and g _2, respectively, may be used, as will be described later. For simplicity of illustration, filters 17 to 19 are shown in broken lines in FIG.
has been omitted.

In the circuit shown in Figure 3a, terminals A and B
The two channels C ₁ and C ₂ connected in between are completely separated, as in FIG. The difference between the two circuits in FIG. 3a and FIG. 1 is that instead of providing the variable gain amplifier in the first path between the device 9 and the output terminal B shared by both paths,
The point is that it is set between The first path variable gain amplifier placed in this new position is designated 11' and its gain is g ₁ '. This amplifier is controlled in exactly the same way as the amplifier 11 of FIG. 1, and everything described for the device of FIG. 1 with respect to the control of the Lutsen oscillation is also valid for the device of FIG. 3a. . In particular, the Lassen oscillations are suppressed to a small non-nuisance amplitude at the output terminal B with a constant value V _L =W _p /A ₂ . Here W _p is the maximum constant value of the signal W supplied by the variable gain amplifier 12 to the input terminal of the linear regulator 13, as explained with reference to FIG.

In another variant, not shown, it is possible to combine the devices shown in FIGS. 1 and 3a,
In that case, variable gain amplifiers 11 and 11' are provided on both sides of device 9 in the first path, respectively, and these are controlled by signals processed by regulator 13.

In the variant shown in FIG. 3b, the two passages C ₁ and C ₂ connected between terminals A and B are not completely separated. This circuit differs from the circuit of FIG. 3a in that the input terminal of the amplifier 14 in the second path is connected to the output terminal of the variable gain amplifier 11' instead of to terminal A, and this terminal A is always connected to the output terminal of the variable gain amplifier 11'. If considered as an input terminal shared by paths C ₁ and C ₂ , amplifier 11' is also shared by both paths, but
The path from the output terminal of this amplifier 11' to terminal B each has a separate path section. If measures are taken to make the gain of the second path higher than the gain of the first path in these individual sections, the Lassen oscillation will occur only in the portion of the second path that is separated from the first path; The amplitude of the oscillation is limited to a constant value V _L =W _p /A ₂ at terminal B, which is small and not a nuisance. In one variant, not shown in the figure, the linear regulator 13 provides a variable gain amplifier (not shown) between the output terminal of the device 9 and the terminal B in addition to the variable gain amplifiers 12 and 11'. A gain amplifier may also be controlled.

The device shown in FIG. 3c can be considered as another improvement of the device in FIG. 3b, since it is possible to dispense with a single variable gain amplifier. The difference is that the second path amplifier 12 is not controlled by the regulator 13 and therefore has a constant gain g ₂ . Considering that in the device of FIG. 3c, the amplifier 11' with a gain g ₁ ' forms part of the first path as well as the second path between the terminals A and B, the regulator 13 The gain g ₁ ' of the amplifier 11' is affected by the amplification links 11', 14, 1 of the second path.
The signal W formed at the output terminal of 2 is held constant.
W _p and feeds this signal to the input terminal of regulator 13. The amplitude of the oscillation at terminal B is still low and held at a constant value W _p /A ₂ .

In this variant of the arrangement of FIG. 3c, the regulator 13 can in particular control a variable gain amplifier, not shown, which is provided between the output terminal of the device 9 and the terminal B.

In the circuit shown in Fig. 3d, the two terminals A
and _{B are} not completely separated. The circuit in FIG. 3d differs from the circuit in FIG. 1 in that an attenuator with a gain of 1/A ₂ is connected between the output terminal of the amplifier 14 and the input terminal of the variable gain amplifier 11. It is in. If this terminal B is considered as an output terminal shared by both paths _C1 and _C2 , these two paths are connected to the amplifier 11.
have in common. Separate parts of the two passages containing the device 9 for the first passage C ₁ and the second passage C ₂
a separate part including an attenuation circuit 15 for terminal A
and the input terminals of the amplifier 11. 3rd d
It should be noted in the figure that the amplifier 12 is not integrated into the second path. A second passage section separate from the first passage includes an amplifier 14 and an attenuation circuit 1.
5, a signal is formed at a point between
Supplied to the input terminal of This regulator 13 controls the gain g ₂ of the amplifier 12 and the gain g ₁ of the amplifier 11 so that the signal W supplied from the amplifier 12 to the input terminal of the regulator 13 is maintained at a constant value W _p .
Since the gains g ₁ and g ₂ of amplifiers 11 and 12 are controlled by the same signal, the amplitude of the Ratsen oscillation at terminal B is limited to a constant value of V _L = W _p /A ₂ · g ₁ / g ₂ It is easy to see that In particular, if the amplifiers are the same, g ₁ =g ₂ and the limit value of the amplitude of the Lutsen oscillation is V _L =W _p /A ₂ .

In this modification of the circuit of FIG. 3d, a variable gain amplifier is provided between terminal A and the input terminal of device 9,
This may be controlled by the regulator 13.

The circuit of FIG. 3e is another improvement of the circuit of FIG. 3d, which corresponds to the improvement of the circuit of FIG. 3c relative to the circuit of FIG. 3b. The circuit of FIG. 3e differs from the circuit of FIG. 3d in that the regulator 13 no longer controls the amplifier 12, but maintains the gain _g2 constant. This regulator 13 acts on the gain g ₁ of the amplifier 11, considered as part of the second path, so that the signal developed in the second path between the amplifier 14 and the attenuation circuit 15 has a constant gain g ₂ After being amplified by the amplifier 12, it is supplied to the input terminal of the regulator with a constant amplitude W _p . In the circuit of Figure 3e,
The Ratsen oscillation amplitude at terminal B is V _L =W _p /A ₂ g ₁ /g ₂ . However, unlike the variant mentioned above, this value
V _L can vary and the gain g ₂ is constant. Nevertheless, the damping factor A ₂ and the gain g ₂ can be chosen such that the variable amplitude of the Lutsen oscillation is kept low and non-nuisance. The variant shown in FIG. 3e is such that an amplifier 11 with a variable gain g ₁ is arranged in the path of the useful audio signal coming from the telephone line and to be fed to a loudspeaker integrated in the electroacoustic circuit 16. Depending on the terminal B
The ratio between the effective audio signal and the noise generated by the non-nuisance Luthsen oscillation caused by the second path can be constant.
Although not shown, in a modification of the circuit of FIG. 3e, a variable gain amplifier may be provided between terminal A and device 9 and controlled by regulator 13.

The circuit shown in FIG. 3f differs from the circuit shown in _FIG . and amplifier 14 to amplifier 1
1' and device 9.
Thus, in the circuit of FIG. 3f, variable gain amplifiers 11' and 11 connected to terminals A and B are used in common for the two paths and are controlled by the same signal supplied from regulator 13. Do what you want. The separate parts of these two paths have the same components as the circuit of _FIG .
It is similarly connected to the second passage. In the circuit shown in Figure 3f, the value of the amplitude of the Ratsen oscillation at terminal B is
V _L =W _p /A ₂ g ₁ /g ₂ and changes in the same way as g ₁ .

In the circuits of FIGS. 3c, 3e, and 3f,
Although we have assumed that the gain g ₂ of amplifier 12 is constant, it is possible to omit this amplifier completely if it is sufficient that the gain g ₂ is equal to 1 for the circuit to operate correctly. It is.

In the circuit shown in FIGS. 3a-3f, it is advantageous to include filters having the same characteristics and functions as the filters 17-19 of the circuit shown in FIG. In that case, a filter having the function of filter 17 is provided in a part of the second passage separated from the first passage, and a filter having the function of filter 18 is provided in a part of the first passage separated from the second passage. A filter having the function of filter 19 is provided in a part common to the main loop and the auxiliary loop.

In the above description, the present invention is used in an electroacoustic system in which loops are formed at every corner by acoustic and electrical coupling. However, the invention is also suitable for general use in all systems in which loops may be formed where high amplitude uncontrolled oscillations may occur. Therefore, the invention can also be used in all types of control systems, such as the one shown in FIG.

FIG. 4 is a block diagram of a known circuit for a control system in which an arbitrary output quantity S is made responsive to an electrical input signal E. To facilitate understanding of this circuit, components and quantities are indicated using the symbols and conventions shown in FIG. Input signal E
is supplied via point (or terminal) A to the forward path of this control system comprising a cascade of a device 9 with a gain G ₁ , a variable gain amplifier 11 and a device 20 providing an output quantity S. A return path of this control system is connected between device 20 and point A, and this return path is connected to a transfer coefficient (transfer coefficient).
function modulus) G circuit 16. In response to this output quantity S, this circuit 16 produces an electrical signal U and adds it to the input signal E with the appropriate phase. During adjustment of this control system or during operation under an abnormal environment, undesired oscillations may occur in the loop formed by the forward path and the return path, or the output quantity S may be uncontrolled and the amplitude may be abnormal. It is also possible that the price may rise.

According to the invention, it is possible to avoid this type of loop oscillation. For example, when using the embodiment shown in FIG _. 1, according to one of the variations of _FIG . A second passage C ₂ comprising an attenuator 15 of /A ₂ is connected to the terminal A of the first passage C ₁ of the control loop comprising components 9 and 11.
The present invention is utilized by connecting to and B.
In this case, the output signal of the amplifier 12 is
3, it is possible to make the gain g ₁ of the amplifier 11 responsive to the gain g ₂ of the amplifier 12 with this regulator. The explanation given for the electroacoustic system shown in FIG. 1 also applies to this embodiment. In particular, possible loop oscillations pass through a second path C ₂ in which the oscillation amplitude is controlled by regulator 13 . Thanks to the damping circuit 15, the amplitude of these oscillations is reduced to the output S of the control system.
The value is extremely low and does not become a nuisance.

FIG. 5 shows an embodiment of a combinational circuit formed by variable gain amplifier 12 and its adjuster 13 and variable gain amplifier 11. In FIG. This embodiment utilizes the principle of a regulating device which is described in detail in French patent application no. 8007055 in the name of the applicant.

According to FIG. 5, the amplifier 12 provided in the second path _C2 comprises an npn transistor, the emitter of which is connected to the negative supply terminal serving as a reference terminal, and the collector connected to a series circuit of a resistor 23 and a capacitor 24. The base is connected to the output terminal of the integrating circuit 25. As will be explained below, this integrating circuit 25 receives the pulsed signal P _c generated in the regulator 13 . This transistor 22 is the adjustable component of amplifier 12. In fact, resistance 2
3 and the capacitor 24, since its level depends more or less on the conduction state of the emitter-collector path of the transistor 22 and thus on the output voltage of the integrating circuit 25.
It is a portion of the input signal x of the amplifier and is variable. As can be readily seen, as the output voltage of integrating circuit 25 increases or decreases, the level of the signal available at terminal 26 also decreases or increases correspondingly. This variable level signal obtained at the terminal 26 is, for example, a current _in , which is supplied to a constant gain amplifier 27. The output terminal 28 of this amplifier 27 has a variable current I _n equivalent to the output signal W of the variable gain amplifier 12.
will occur. Further, a current I _p +I _n is generated at the output terminal 29 of the amplifier 27, where I _p is a direct current with a constant amplitude. In this regulator 13, the current I _p +
I _n is supplied to a pulse width modulator 30 . This modulator 30 receives clock pulses from a clock generator 31 and produces a signal P _n formed of pulses whose width is modulated by a current I _p +I _n . Methods of forming pulse width modulated pulses are known. Here, in particular, clock pulses are used to sample the input current I _p +I _n of the modulator. If the variable portion I _n of this current at the time of sampling is zero, the width of the pulse of the signal P _n is P _p . Depending on whether the variable current I _n is positive or negative at these sampling instants,
The pulse width of signal P _n becomes longer or shorter than P _p . Pulse length P _p + P _n1 and P _p −
A pulse with P _n1 has a variable current I _n of values I _n1 and −I _n1
This corresponds to a predetermined level of this current I _n when the current I n can be taken. This modulated signal P _n is provided to a pulse length detector 32, hereinafter commonly referred to as a compressed pulse, having a constant duration and a signal P _n
A pulse-shaped pulse-shaped signal P _c is generated each time a pulse of P c reaches the value P _p ±P _n1 .
The pulse shaped signal P _c is applied to an integrator circuit 25 which produces a voltage representing the pulse repetition rate of the compressed pulse. Basically, this integrator circuit comprises a capacitor that is charged with a constant current during the duration of the compression pulse and discharged with a current that is lower than this charging current. Thus, when a compression pulse occurs representing the occurrence of a level of current I _n , the voltage developed from integrator circuit 25 increases, causing transistor 22 to become more conductive and attenuating the current level I _n . Eventually, the voltage generated from the integrating circuit 25 stabilizes around the average value, making the current I _n at a constant level. According to the embodiment described above, the amplifier 12 associated with the _regulator 13 has exactly the required characteristics shown in _FIG . is equal to the constant level.

Variable gain amplifier 11 incorporated in the first path C ₁
has the same configuration as the amplifier 12, therefore,
The components are indicated by the same number with a dash '. The amplifier 11 is controlled by the pulsed signal P _c generated by the regulator 13 described above. Transistor 22', which is a variable component of amplifier 11, is controlled by the same signal as transistor 22, which is a variable component of amplifier 12. The two amplifier components in particular transistors 22 and 22'
are arranged in pairs in a suitable manner, the gain of amplifier 11 is changed at each moment to that of amplifier 12.
The gains of both can be automatically adjusted to the same value.

Since the purpose of the amplifier 11 is to process a useful signal (e.g. an audio signal), the amplifier 1
2 is to process the oscillating signal of the loop, it may be advantageous to provide the integrator circuits 25 and 25' with separate time constants to suit the signals processed by this amplifier. On the other hand, the transistor 22
and 22' are arranged as a pair and controlled by the same signal, thereby giving different gains to amplifiers 27 and 27' and maintaining the gains of amplifiers 11 and 12 at a constant ratio.

The circuit shown in FIG. 5 is the circuit shown in FIG. 1 and the amplifier 11.
12 and a regulator 13. FIG. However, in the same embodiment, the regulator 13 controls two variable amplifiers that are otherwise connected, but only one of them, resulting in several variations shown in FIG. It is clear that this can be realized.

When the present invention is applied to a hands-free type telephone set for avoiding non-nuisance lassen oscillations, it can be very easily made to perform even more significantly useful functions. That is, complementary gain changes can be made in the transmit and receive paths of the telephone.

In the case of known telephones, in order to achieve complementary gain changes in the two paths, a unidirectional gain change is performed with the aid of an amplifier in the transmit path and at the same time with the aid of an amplifier in the receive path. It is necessary to change the gain in the other direction, although with the same amplitude. To perform these gain switches by sound, two sound detectors are often used whose output signals are used to perform the gain switches.

The present invention allows complementary gain changes in the two paths of the telephone to be obtained by controlling the gain of only one single amplifier. In order to explain this feature according to the invention, consider the case of a telephone associated with the device according to the invention as shown in FIG. 1, and the gain switching method will be explained with reference to FIG. In FIG. 6, most of the components of the circuit shown in FIG. 1 are designated by the same reference numerals. In this embodiment, the single amplifier that serves to control the gain of the two paths of the telephone is amplifier 5, which in this case has variable gain and which is connected to the transmit path. , microphone, first passage C ₁ and second passage
It shall be inserted between the input terminal A and the input terminal A, which is jointly used by _C2 . When the amplifier 5 has a minimum gain value, that is, the gain G is a minimum value, the second path
If the gains of the components of C ₂ are controlled as described above so that non-nuisance connection oscillations occur only in the auxiliary loop formed by circuit 16 and the second path, these non-nuisance oscillations are removed by amplifier 5. The gain g
occurs reliably for all other values of . gain g
For all values of , the total gain of the auxiliary loop is given by Equation (7)
remains equal to 1, expressed as .

Since the amplitude of the oscillation passing through the circuit 16 is very small, the components of this circuit operate in a linear mode and the gain G of this circuit 16 is approximately proportional to the gain g.
If h is a constant coefficient and G=hg, equation (3) may be rewritten as h・g・g ₂・G ₂・1/A ₂ =1 (9).

If two amplifiers 11 and 12 with variable gains g ₁ and g ₂ are equivalent and controlled with the same sign, then g ₁ = g ₂ and equation (9) becomes h・g・g ₁・G It can be rewritten as ₂・1/A ₂ =1 (10).

From this, the gain g ₁ of the amplifier 11 that amplifies the signal received by the telephone corresponds to the change in each gain g of the amplifier 5 that amplifies the signal from the microphone 4.
automatically change with the same amplitude and in opposite directions.

When the gains g ₁ and g ₂ are controlled so that g ₁ = ag ₂ , the gain in the receiving path changes in the opposite direction and with an amplitude of 1/a in response to a change in each gain g of the amplifier 5. It is easy to see that it changes.

It should be noted that in order to reverse the gain changes in _the transmit and receive paths, an amplifier, _e.g. Amplifier 5 needs to be activated. A gain change occurring anywhere in the receive path, such as a gain change in amplifier 7, does not cause a gain change in the transmit path.

The gain of the amplifier 5 may be changed gradually or abruptly between two gain values. In order to automatically switch the voice of the telephone, it can also be controlled, for example, by the voice signal detector 40. When a subscriber speaks into microphone 4, the voice detected by detector 40 acts to control the normal gain of amplifier 5. Therefore, the gain g ₁ for the amplifier 11 automatically becomes a small value, and therefore the energizing signal of the speaker 8 also becomes weak. When the subscriber stops speaking into the microphone 4, the low gain of the amplifier 5 is controlled by the detector 40, and the gain _g1 of the amplifier 5 is automatically set to a large value, and the normal energization signal is sent to the speaker 8. will be supplied. In this case, the subscriber may also exercise manual control to achieve the best listening conditions.

In addition, switching the gain in the first and second paths,
It is easily understood that other variations of the invention can also be implemented using a single variable gain amplifier. A variable gain amplifier in the first path _C1 , controlled by a regulator 13, is connected to the telephone 2
By providing in one of the two paths a variable gain amplifier used for gain switching between the acoustic transducer in the other path and the terminals shared by the two individual parts of both paths C ₁ and C ₂ . It is sufficient to provide it in between. For example, in the device shown in FIG. 3a, a variable gain amplifier 11' controlled by a regulator 13
is provided in the transmission path of the telephone, and a variable gain amplifier used for gain switching is installed between the output terminal B shared by both paths and the speaker incorporated in the electroacoustic circuit 16. In the receiving passage of
It may be provided. In order to automatically switch the gain depending on the audio signal, it is also possible to employ a detector 40 that detects the audio signal of the microphone. What should _be noted here is that the regulator ₁₃
In devices such as those shown in FIGS. 3c, 3e, and 3f, which control only one or two variable gain amplifiers shared by the Variations can be obtained by means of gain switching amplifiers placed in appropriate paths.

It is clear that the invention is not limited only to the embodiments described above, but can be subjected to many variations and modifications.

[Brief explanation of drawings]

Fig. 1 is a block diagram showing an embodiment of the device of the present invention used in a public address telephone, Fig. 2 is a characteristic diagram showing the characteristics of a regulator used in the device of the present invention, and Fig. 3a
3F are block diagrams showing modified examples of the present invention, FIG. 4 is a diagram showing an embodiment of the device of the present invention used in a control system, and FIG. 5 is a diagram showing a regulator,
FIG. 6 is a diagram illustrating an embodiment of the invention with two variable gain amplifiers controlled thereby; FIG. 6 shows the arrangement of FIG. FIG. 1... Coupling circuit, 2... Telephone line, 3... Transmission path (talking path), 4... Microphone, 5, 14,
27, 27'...Amplifier, 6...Reception path (reception path), 7...Reception amplifier, 8...Speaker (loudspeaker), 9, 20...Device, 10...Circuit,
11, 11'... variable gain amplifier, 12... another variable gain amplifier, 13... linear regulator, 15...
Attenuation circuit, 16... Electroacoustic circuit, 17, 18,
19... Filter, 22, 22'... Transistor, 23, 23'... Resistor, 24, 24'... Capacitor, 25, 25'... Integrating circuit, 30... Pulse width modulator, 31... Clock Generator, 32...
...Pulse length detector, 40...Audio signal detector.

Claims (1)

[Claims] 1. A first passage having an input terminal and an output terminal;
A circuit for preventing undesired oscillations occurring within a predetermined frequency band in a closed loop system including at least a feedback path between the input and output terminals of the first path, the circuit including a variable gain amplifier between the input and output terminals of the first path. In the built-in undesired oscillation prevention circuit, a second passage, at least a portion of which is separated from the first passage, is provided between the input terminal and the output terminal of the first passage, and further includes a second passage within the second passage. at least one variable gain amplifier incorporated therein; means for maintaining the gain of the second path higher than the gain of the first path over the frequency band in which oscillations are likely to occur; a linear regulator having an input terminal for receiving a signal formed in the second path portion and an output terminal connected to a control terminal of the at least one variable gain amplifier forming part of the second path. , the linear regulator controls the at least one variable gain amplifier with a signal at the input terminal of the linear regulator,
That is, an undesired oscillation prevention circuit characterized in that the signal in the second path is controlled so as to be maintained constant from a predetermined value. 2. The second path is completely separated from the first path, and the second path is provided with a variable gain amplifier for producing an output signal that is applied to the input terminal of the linear regulator, whereby the linear regulator While controlling the variable gain amplifier of the second path, a variable gain amplifier of the first path provided near an input terminal shared by these first and second paths; 2. The undesired oscillation prevention circuit according to claim 1, wherein the undesired oscillation prevention circuit is configured to control both or one of the variable gain amplifier and the variable gain amplifier provided near the output terminal shared by the second path. 3. The undesired oscillation prevention circuit according to claim 2, wherein the second path is connected to the output terminal shared by the first and second paths by an attenuation circuit. 4. The first and second paths include one common variable gain amplifier provided near the input terminal shared by both these paths and a variable gain amplifier provided near the output terminal shared by both these paths, or Unwanted oscillation according to claim 1, characterized in that one or the other of these variable gain amplifiers shared by both paths is controlled by the linear adjuster. prevention circuit. 5. The first and second paths include a common single variable gain amplifier located near an input terminal shared by both paths (or an output terminal shared by both paths). In the undesired oscillation prevention circuit described above, the first path is provided near an output terminal shared by both the paths (or an input terminal shared by both paths) and controlled by the linear regulator. An undesired oscillation prevention circuit comprising another variable gain amplifier. 6. Claims characterized in that the signal formed in the portion of the second path separated from the first path is configured to be supplied to the input terminal of the linear regulator by a constant gain amplifier. 4. The undesired oscillation prevention circuit according to 4 or 5. 7 configured to supply a signal formed in a portion of the second path separated from the first path to an input terminal of the linear regulator by a variable gain amplifier controlled by the linear regulator; An undesired oscillation prevention circuit according to claim 4 or 5, characterized in that: 8. Claims 4 to 7, wherein said first and second paths include one common single variable gain amplifier located near an input terminal shared by both paths.
2. The undesired oscillation prevention circuit according to any one of the above, characterized in that the second path is connected to an output terminal shared by both paths by an attenuation circuit. 9. Unwanted oscillation according to any one of claims 4 to 7, wherein the first and second paths include a common single variable gain amplifier located near an output terminal shared by both paths. An undesired oscillation prevention circuit characterized in that a portion of the second path separated from the first path is connected to an input terminal of the common variable gain amplifier by an attenuation circuit. 10. Claims 4, 6 and 7, wherein the first and second paths include two shared variable gain amplifiers respectively located near the input and output terminals shared by both paths. In the undesired oscillation prevention circuit according to any one of the above, a portion of the second path separated from the first path is provided near an output terminal shared by both the paths by an attenuation circuit. An undesired oscillation prevention circuit characterized in that the undesired oscillation prevention circuit is connected to an input terminal of a common variable amplifier. 11 The means for maintaining the gain of the second path higher than the gain of the first path over a frequency band where oscillations are likely to occur is a portion of the second path separated from the first path, and is a linear regulator. Claims 1 to 10 characterized by comprising a filter disposed upstream of a point where the signal supplied to the input terminal is formed and selectively increases the gain of the second path in the frequency band. The undesired oscillation prevention circuit described in any one of the above. 12. Claim 11, wherein the filter in the second passage is a narrow band filter.
The undesired oscillation prevention circuit described above. 13. The undesired oscillation prevention circuit according to claim 11, wherein the filter in the second path is a high-pass filter. 14 The means for maintaining the gain of the second path higher than the gain of the first path over a frequency band in which oscillation is likely to occur is arranged in a portion of the first path separated from the second path, and Claims 11 to 12 further include a filter for selectively increasing the gain of the first passage in the
13. The undesired oscillation prevention circuit according to any one of 13. 15 The means for maintaining the gain of the second path higher than the gain of the first path over a frequency band where oscillation is likely to occur is provided in both the paths, in front of the input terminal shared by the first and second paths. The present invention is characterized by comprising a filter disposed after the shared output terminal and within a portion shared by both the paths, for selectively increasing the gain of the combinational circuit of both paths in the frequency band. Claims 11-1
3. The undesired oscillation prevention circuit according to any one of 3. 16 The linear regulator comprises a pulse width modulator receiving a signal corresponding to a signal applied to an input terminal of the linear regulator, and the width of the pulse produced by the pulse width modulator increases the signal level at its input terminal. a pulse length detection circuit that generates a compression pulse each time a value representing that . 16. The undesired oscillation prevention circuit according to claim 1, wherein the circuit is configured to control variable components of one or more variable gain amplifiers.